Soil Ecotoxicology Needs Robust Biomarkers: A Meta-Analysis Approach to Test the Robustness of Gene Expression-Based Biomarkers for Measuring Chemical Exposure Effects in Soil Invertebrates

Abstract

Gene expression-based biomarkers are regularly proposed as rapid, sensitive, and mechanistically informative tools to identify whether soil invertebrates experience adverse effects due to chemical exposure. However, before biomarkers could be deployed within diagnostic studies, systematic evidence of the robustness of such biomarkers to detect effects is needed. In our study, we present an approach for conducting a meta-analysis of the robustness of gene expression-based biomarkers in soil invertebrates. The approach was developed and trialed for two measurements of gene expression commonly proposed as biomarkers in soil ecotoxicology: earthworm metallothionein (MT) gene expression for metals and earthworm heat shock protein 70 (HSP70) gene expression for organic chemicals. We collected 294 unique gene expression data points from the literature and used linear mixed-effect models to assess concentration, exposure duration, and species effects on the quantified response. The meta-analysis showed that the expression of earthworm MT was strongly metal concentration dependent, stable over time and species independent. The metal concentration-dependent response was strongest for cadmium, indicating that this gene is a suitable biomarker for this metal. For copper, no clear concentration-dependent response of MT gene expression in earthworms was found, indicating MT is not a reliable biomarker for this metal. For HSP70, overall marginal up-regulation and lack of a concentration-dependent response indicated that this gene is not suitable as a biomarker for organic pollutant effects in earthworms. The present study demonstrates how meta-analysis can be used to assess the status of biomarkers. We encourage colleagues to apply this open-access approach to other biomarkers, as such quantitative assessment is a prerequisite to ensuring that the suitability and limitations of proposed biomarkers are known and stated. Environ Toxicol Chem 2022;41:2124-2138. © 2022 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.

Keywords: Biomarker; Gene expression; Heat shock protein; Meta-analysis; Metallothionein; Soil invertebrate.

Figure 1

Overview of the evidence collection methods and results. (A) Evidence identification map with the numbers of publications remaining and removed at each stage. (B) The total cumulative number of relevant publications per year (closed circles) and the number of relevant papers per year (bars). WoS, Web of Science.

Figure 2

Overview of the composition of the species biomarker dataset showing the number of studies per (A) study type, (B) species and species group, (C) metal, (D) organic pollutant class, (E) omics method used for expression measureent, and (F) biomarker. For heat shock proteins (HSPs), the number shown is the total number of studies that use any class of HSP (e.g., HSP70, HSP90, etc.). ABC, ABC transporters; CAT, catalase; CYP, cytochrome p450; GST, glutathione‐S‐transferase; HSP70, heat shock protein 70; MT, metallothionein; PAH, polycyclic aromatic hydrocarbon; qPCR, quantitative polymerase chain reaction; SOD, superoxide dismutase.

Figure 3

Number of relevant experimental studies that used earthworms per biomarker and chemical group. ABC = ABC transporters; CAT = catalase; CYP = cytochrome p450; GST = glutathione‐s‐transferase; HSP70 = heat shock protein 70; MT = metallothionein; SOD = superoxide dismutase.

Figure 4

Relationship between log2 fold change of metallothionein (MT) gene expression in earthworms and (A) soil metal concentration, (B) exposure duration, (C) soil metal concentration per metal, (D) exposure duration per metal, (E) metal, (F) study type, (G) earthworm species, and (H) soil metal concentration per species. In (E)–(H), open circles represent data points, and lower and upper boxes correspond to the 25th and 75th percentiles with the upper/lower whiskers extend from the box to the highest/smallest value at most 1.5 × interquartile range of the box. Note that the different data points in the graphs are not all independent from each other but nested in “Publication.” Therefore, patterns (or in some cases the lack of) that can be observed through visual inspection do not always match statistical results. In (F): spiked, spiked experiments; field, polluted field soil experiments. In (G)–(H): Ac = Aporrectodea caliginosa; Ea = Eisenia andrei; Ef = Eisenia fetida; Lr = Lumbricus rubellus; Lt = Lumbricus terrestris.

Figure 5

Relationship between log2 fold change of heat shock protein 70 (HSP70) gene expression in earthworms and (A) the soil concentration of organic pollutants, (B) mechanism of action, (C) exposure duration, and (D) the interaction between exposure duration and mechanism of action. In (B), open circles represent data points, and lower and upper boxes correspond to the 25th and 75th percentiles with the upper/lower whiskers extending from the box to the highest/smallest value at most 1.5 × interquartile range of the box. GABA = gamma‐aminobutyric acid gated chloride channel; nAChR = nicotinic acetylcholine receptor. Note that in (D), not all coefficients (sublevels) of the interaction between exposure duration and mechanism of action are shown as most of these sublevels did not have enough datapoints to assess interaction effects.